These are both more rare than car-bicycle collisions and less likely to be injurious.

> also exclusive roads for bicycles are usually of low quality not up to engineering standards for roads, they will be full of dirt, unmaintained, with potholes

Citation needed? The main problem I've experienced on bike paths is tree roots. Roads for bikes don't wear out particularly quickly, because bikes don't make anything like the kind of wear and tear on roads that cars or especially larger vehicles (buses and trucks) do. Road wear is proportional to weight squared.

> Most bicycle accidents are not even collisions with cars

What are they collisions with? Do you think bike-bike, bike-ped, and bike-fixed object collisions combined are more frequent than bike-car collisions? I don't have numbers on this one way or another. To provide some color, motor vehicle-fixed object collisions are about a third of all motor vehicle crashes[0].

Additionally, 95% of cyclist collision deaths are definitely car-bike collisions[1]. So even if they are not the majority of collisions, they must be disproportionately deadly and worth consideration anyway.

[0]: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/... p. 18 (table 5(b)).

[1]: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/... p. 5-6.